Artificial intelligence tools can be used to summarize a book, predict what a shopper might like, guide an autonomous vehicle, and analyze medical images. But Robert Krueger wants to tap the technology to do something unexpected that could be critical for millions of people around the world: identify solutions to poverty.
Krueger, professor and head of the Department of Social Science and Policy Studies, is launching a three-year project that will use a branch of AI called machine learning to comb through data collected in three countries by Fundación Paraguaya. The organization created and manages a tool called Poverty Stoplight. Krueger’s goal is to find “clusters” where specific economic activities could be established to benefit individuals and communities.
“It’s important to understand the needs of people in low- and middle-income countries, the resources they possess, and what they say they need to gain self-sufficiency through economic development,” Krueger says. “With AI, we can rapidly parse through large amounts of data to identify opportunities, not just needs.”
What makes Krueger’s project different is its reliance on data that comes from the perspective of local communities. The data from Poverty Stoplight is critical, Krueger says, because the organization focuses on capturing how families define their own economic challenges. That approach differs from economic analyses that define poverty by a single factor, such as income.
When asked about their economic challenges, Krueger says, people might link their poverty not just to income, but to multiple factors or personal problems, such as a leaky roof that needs repair. In the same way, economic responses to address poverty will differ from community to community.
“In a community that lacks good sanitation, training masons and carpenters to construct composting toilets could provide jobs to individuals and services that enable families to improve their economic situations,” Krueger says. “Market opportunities are solutions rooted in community needs.”
The project is supported by a $247,250 grant from the Templeton World Charity Foundation. Krueger and a graduate student will develop machine learning algorithms, which teach a computer to recognize patterns and make decisions, to analyze data from Paraguay, Kenya, and Egypt. Then the researchers will verify their findings by making observations and conducting interviews in communities that participated in Poverty Stoplight surveys.
The project builds on Krueger’s research and teaching on sustainability and social justice in global development. He is the director of WPI’s Institute of Science and Technology for Development and leads the Integrated Global STEM Publishing and Education Initiative, which was launched in 2023 by WPI and independent academic publisher De Gruyter.
Krueger says a secondary goal of his new project is to show how the rapidly growing field of AI can be informed by social scientists and used to improve policies and their interventions. That means showing data scientists how their expertise can be used in new ways and demonstrating to social scientists that the tools of AI are potentially positive, Krueger says. WPI has taken an interdisciplinary approach to the social sciences and technology since the founding of the Social Science and Policy Studies department 50 years ago.
“The potential of AI and the social sciences lies in how we use information and apply our different skills to pull correlations and explanations out of vast amounts of data so we can take the critical steps that will be necessary to address humanity’s most pressing problems,” Krueger says.
